I just want to check an assumption before I go diving off into code; the standard Vertex Texture Fetch (text2dlod) with "manual" linear interpolation as shown here http://www.catalinzima.com/tutorials/4-uses-of-vtf/terrain-rendering-using-heightmaps/ wont work with a GeoClipMap style of heightmap rendering will it ?
I have an initial version which gives very weird results, and I'm attributing it to the fact that the point sampler used in text2dlod is a nearest point sampler, but linear interpolation examples that I've seen for VTF show a simple frac() of the initial texture coords to get the weighting for the neighbour texture pixels. I think this wont be accurate will it ?; becuase the frac() is carried out on the initial texture coordinates, but the point sampler isn't working on a trunc() basis but a round() basis - if the geometry is fixed to that texture coord then you'd never notice it, but with a GeoClipMap the texture coord is calculated from world space dynamically and constantly changes - which means I get weird jumping artefacts as the camera moves and the vertexes texture coord slides from one pixel to another - suddenly jumping pixels when the rounding point is hit (I presume also that its Bankers Rounding so actually behaves differently for each pixel offset).
Can anyone with a geoclipmap implementation tell me if this is the right direction I'm going it, or am I chasing shadows ?